update 新增热点帖子抽取存库脚本

crawler_sys/scheduler/cal_ni_and_put_to_backend.py

@@ -247,14 +247,18 @@ def push_data_to_user(res_data: Dict) -> Dict:
     :return:
     """
     qiniu_img_list = []
+    content = ""
     if res_data["img_list"]:
         for img_url in res_data["img_list"]:
+            try:
                 img_wb = retry_get_url(img_url).content
                 res = upload(img_wb)
                 print(res)
                 img_info = retry_get_url(res + "-imageinfo")
                 img_info_json = img_info.json()
-            qiniu_img_list.append((res, img_info_json))
+                qiniu_img_list.append((res+"-w", img_info_json))
+            except:
+                return {}
 
         # 替换图片
         if res_data["platform"] == "weibo":
@@ -263,8 +267,10 @@ def push_data_to_user(res_data: Dict) -> Dict:
             content = res_data.get("content")
             if content:
                 for count, img_url in enumerate(res_data["img_list"]):
-                    content.replace(img_url, qiniu_img_list[count][0])
+                    # print(qiniu_img_list[count][0])
+                    content = content.replace(img_url, qiniu_img_list[count][0])
                 res_data["qiniu_img_list"] = qiniu_img_list
+                res_data["content"] = content
     if res_data["platform"] == "weibo":
         res_data["content"] = gm_convert_html_tags(res_data["title"], all_tags=True)
         res_data["title"] = ""
@@ -297,7 +303,7 @@ img_type = {
 def write_data_into_mysql(res_data):
     now_str = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
     data = push_data_to_user(res_data)
-    if data.get("title"):
+    try:
         sql_query = """insert into api_tractate 
             (user_id,content,is_online,status,platform,content_level,is_excellent,create_time,last_modified,user_del,low_quality,low_quality_deal,platform_id,pgc_type,title) 
             values ({user_id},'{content}',{is_online},{status},{platform},{content_level},{is_excellent},'{create_time}','{last_modified}',{user_del},{low_quality},{low_quality_deal},'{platform_id}',{pgc_type},'{title}');""".format(
@@ -306,8 +312,6 @@ def write_data_into_mysql(res_data):
         is_excellent=0, create_time=now_str,
         last_modified=now_str, user_del=0,
         low_quality=0, low_quality_deal=0, platform_id=data["doc_id"], pgc_type=1, title=data["title"])
-
-    try:
         res = cur.execute(sql_query)
         tractate_id = int(conn.insert_id())
         if res:
@@ -322,15 +326,38 @@ def write_data_into_mysql(res_data):
             else:
                 image_url_source = 1
             try:
-                image_type = img_type.get(img_info[0].split(".")[-1].upper())
+                image_type = img_type.get(img_info[1]["format"].upper())
             except:
                 image_type = 1
+            try:
+                width = img_info[1]["width"]
+                height = img_info[1]["height"]
+            except:
+                width = 0
+                height = 0
+            try:
+                if img_type == 7:
                     sql_query = """
-                insert into api_tractate_images (tractate_id,image_url,width,height,image_url_source,image_type,image_webp,create_time,update_time)
-                values ({tractate_id},{image_url},{width},{height},{image_url_source},{image_type},{image_webp},'{create_time}','{update_time}')
-                """.format(tractate_id=tractate_id, image_url=img_info[0], width=img_info[1]["width"],
-                           height=img_info[1]["heigh"], image_url_source=image_url_source, image_type=image_type,
+                                        insert into api_tractate_images (tractate_id,image_url,width,image_webp,height,image_url_source,image_type,image_webp,create_time,update_time)
+                                        values ({tractate_id},'{image_url}',{width},{height},{image_webp},{image_url_source},{image_type},{image_webp},'{create_time}','{update_time}')
+                                        """.format(tractate_id=tractate_id, image_url=img_info[0], width=width,
+                                                   height=height, image_url_source=image_url_source,
+                                                   image_type=image_type,image_webp=img_info[0],
                                                    create_time=now_str, update_time=now_str)
+                else:
+                    sql_query = """
+                    insert into api_tractate_images (tractate_id,image_url,width,height,image_url_source,image_type,create_time,update_time)
+                    values ({tractate_id},'{image_url}',{width},{height},{image_url_source},{image_type},'{create_time}','{update_time}')
+                    """.format(tractate_id=tractate_id, image_url=img_info[0], width=width,
+                               height=height, image_url_source=image_url_source, image_type=image_type,
+                               create_time=now_str, update_time=now_str)
+                res = cur.execute(sql_query)
+
+                if res:
+                    conn.commit()
+            except Exception as e:
+                print("commit error %s" % e)
+                conn.rollback()
 
 
 def task_main():
@@ -344,7 +371,8 @@ def task_main():
     # 循环处理抓取数据,返回需要添加至后端的数据
     for res_data in scan_from_redis(rules_list):
         write_data_into_mysql(res_data)
-
+    # test = {'release_time': 1595952037000, 'fetch_time': 1596012816514, 'url': 'https://www.douban.com/group/topic/186707979/', 'releaser': '🍫', 'repost_count': 40, 'comment_count': 411, 'favorite_count': 144, 'title': '王俊凯终于还是举铁了', 'releaserUrl': 'https://www.douban.com/people/57762442', 'releaser_id_str': 'douban_57762442', 'video_img': 'https://img3.doubanio.com/view/group_topic/sqxs/public/p317684082.webp', 'mid': '186707979', 'platform': 'douban', 'doc_id': 'douban_186707979', 'content': '<div id=\'content\'><div class="image-container image-float-center"><div class="image-wrapper"><img src="https://img3.doubanio.com/view/group_topic/l/public/p317684082.webp" width="500"/></div></div><div class="image-container image-float-center"><div class="image-wrapper"><img src="https://img9.doubanio.com/view/group_topic/l/public/p317684064.webp" width="500"/></div></div><div class="image-container image-float-center"><div class="image-wrapper"><img src="https://img3.doubanio.com/view/group_topic/l/public/p317684093.webp" width="500"/></div></div><div class="image-container image-float-center"><div class="image-wrapper"><img src="https://img9.doubanio.com/view/group_topic/l/public/p317684095.webp" width="500"/></div></div><div class="image-container image-float-center"><div class="image-wrapper"><img src="https://img3.doubanio.com/view/group_topic/l/public/p317684052.webp" width="500"/></div></div><p></p></div>', 'collection_count': 107, 'img_list': ['https://img3.doubanio.com/view/group_topic/l/public/p317684082.webp', 'https://img9.doubanio.com/view/group_topic/l/public/p317684064.webp', 'https://img3.doubanio.com/view/group_topic/l/public/p317684093.webp', 'https://img9.doubanio.com/view/group_topic/l/public/p317684095.webp', 'https://img3.doubanio.com/view/group_topic/l/public/p317684052.webp'], 'level': 5}
+    # write_data_into_mysql(test)
     cur.close()
     conn.close()